Tumors are variable and heterogeneous and therefore could escape immunotherapy targeted to a specific antigen. For an effective cancer treatment, it is important to target the cell population responsible for propagating a tumor and causing tumor progression.
Adoptive cell therapy (ACT) is the transfer of immune cells as a means of enhancing immune functionality. Cytolytic T lymphocytes (CTL) expanded from a patient's tumor infiltrating lymphocytes (TIL) are capable of enhancing the killing of tumor cells to cause regression of the tumor and in some patients, a durable remission effectively curing the patient (Rosenberg, J. Nat. Cancer Inst. 86(15):1159-66 (1994)). However, the use of autologous TIL as a personalized cancer therapy has several drawbacks: 1) adequate numbers of TILs can only be generated in about 40-50% of patients (P. Hwu, Presentation, Society for Immunotherapy of Cancer, 25th Annual Meeting, Oct. 1, 2010), 2) it takes 4-6 weeks to generate enough T cells and 3) the antigens targeted by the TIL are diverse (Nishimura, et al. J. Immunotherapy with Emphasis on Tumor Immunology: Official Journal of the Society for Biological Therapy 16(2):85-94 (1994); Rosenberg, 1994; Shilyansky, Proc. Nat, Acad. Sci. USA 91(7): 2829-33 (1994)), which makes its effectiveness difficult to standardize (Schwartzentruber D J, et al. J Clin Oncol. 12(7):1475-83 (1994)) and ensure that the TIL target clinically meaningful antigens.
One way to control the targeted T cell killing more selectively without requiring expansion of patient TIL is to genetically engineer T cell receptors (TCR) to be specific for single known antigens. ACT using CTL has shown exceptional promise in the treatment of certain non-epithelial and viral-induced cancers. However the paucity of validated antigens has limited the use of engineered T cells to a small number of well-characterized antigens (e.g., gp100, NY ESO-1, WT-1, MART-1, MAGE-3), although they too lack definitive data as to how they relate to the tumor propagating population of the cancers (for example, the cancer testis antigen NY ESO-1 is a potentially important ACT target in epithelial ovarian cancer yet differences in intratumor and intertumor expression of the antigen vary (Woloszynska-Read et al., Clin. Canc. Res, 14:3283 (2008)) and must be better understood for effective cancer immunotherapy using this antigen).
To achieve ACT with curative potential, a way is needed to validate the relationship of presently known and novel antigens to tumor propagation and progression as well as to determine antigen cross-reactivity with normal cells, particularly those undergoing normal wound repair and regeneration. This information would enable the safe and effective targeting of ACT using engineered T cells and expand the clinical application of ACT, particularly to cancers originating from vital organs where the use of tissue-specific markers (e.g. MART-1 against melanocytes) is not an option.
Such a solution is particularly needed to effectively treat cancers arising from endogenous functional mutations (as opposed to viral mutations), which have proven to be particularly challenging. Some believe that ‘shared unique’ tumor antigens would make the ideal target for adoptive engineered T cell therapy and that antigens arising from specific mutations will only identify a small subset of tumors within a certain cancer type and stage (Paschen, T Cell Antigens in Cancer, Chapter 1, Tumor-Associated Antigens, Gires and Seliger eds. Wiley-VCH GmBH & co. KGaA, Weinheim, (2009)). Researchers have postulated that fully cataloging genetic rearrangements and transcriptional changes is a way to overcome this limitation: “Central to expectations of accelerated target discovery is the perception that genome, transcriptome and proteome analyses will lead to the discovery of molecules against which cancer therapeutics might be targeted” (Strausberg et al., Nature 429(6990):469-474 (2004)). Yet oncogenomics is of limited use for curative targeting of engineered T cell therapy because it merely associates the presence of a genetic alteration with the stage of the tumor. While this information can be of prognostic value it does not provide the information necessary to determine the mutation's appropriateness for curative targeting. It is widely appreciated that without being able to link genetic changes to functional biology, genomic targets will not translate to drugs or biomarkers that can be used for highly effective biologic therapeutics such as ACT. An additional practical consideration that confounds ‘omics’ attempts at determining the relevance of biomarkers directly from tissue samples is the significant bias that is introduced with even small changes in tumor sample handling and processing (Ransohoff and Gourlay, J Gin Oncol 28(4):698-704 (2010).
Researchers have also employed biological methods as a way to better understand cancer and identify therapeutic targets. They have recognized that accurate determination of the predictive and therapeutic significance of cancer markers depends on the comparison of expression signatures in normal lineages with those of different tumor subtypes (Visvader, et al. Nature 469(7330):314-322 (2011)). But population heterogeneity and a growing realization that the tumor cell population is both variable and dynamic (Gupta, et al. Nature medicine 15(9):1010-1012 (2009); Gupta, et al. Cell 146(4):633-644 (2011)), has made meaningful comparison difficult.
Despite the heterogeneous nature of a tumor, its histopathology and gene expression can appear relatively stable as it progresses from localized disease to metastatic and end-stage disease (Visvader, et al. Nature 469(7330):314-322 (2011)). Also, tumor phenotype does not necessarily translate to tumor histopathology or lineage marker expression (Visvader, et al. Nature 469(7330):314-322 (2011)). Leukemia and solid tumors can maintain a differentiation hierarchy (Shipitsin, et al. Lab Invest. 88(5):459-63 (2008)), but the relationship within the hierarchy is disrupted, where cancer-propagating cells are similar to progenitor and stem cells in some ways, but different in other ways. Because of this, cancer has been referred to as a “disease of differentiation.” More precisely, cancer is a disease of genes operating in the context of differentiation. Therefore what is needed is a way to simplify cancer's dynamic nature and variability within a context of functional differentiation so that therapeutic targets with curative potential can be identified.
Analysis of fractionated cells, single cells and in situ expression of biomarkers in a static condition are current approaches to identifying tumor subpopulations. However they do not provide a means to establish the relevance of the biomarker to the tumor propagating subpopulation(s), which is needed to ensure curative therapeutic targeting.
Although many have attempted to identify and, select out cells at varying stages of differentiation using known markers, these cell compartments cannot be mechanically selected with either stem cell or differentiation markers and then expanded for analysis because 1) regeneration-capable cells are identified primarily by their behavior and 2) it is not certain to what extent regeneration-capable tumor propagating cells will express stem cell or differentiation markers (Tysnes, Neoplasia 12(7):506-515 (2010)). The regeneration-capable cells of a cancer (Maenhaut, Carcinogenesis 31(2):149-158 (2010)), are referred to herein as the “C-RC”, and encompass all regeneration capable cells regardless of their cell compartment of origin, and may encompass cancer cells with properties of normal stem cells, but need not be limited to or related to a native stem cell compartment. Therefore what is needed is the opposite approach, i.e., a way to first identify and isolate the cell subpopulation based on its function, and then determine its identifying markers.
Researchers have looked to fetal cells as a way to identify stem and progenitor cell markers because fetal cells, in the process of organogenesis, will have a naturally rich generative cell pool (U.S. Pat. No. 7,078,231 to Roberts, et al.). While many cancers may arise in progenitor cells, abnormalities can occur in the stem, progenitor, or transit amplifying cell compartments (Tysnes, Neoplasia 12(7):506-15 (2010)). This makes the tactic of relying on the natural generative pool present in fetal tissue (U.S. Pat. No. 7,078,231 to Roberts, et al.) or derived from adult tissue through regenerative activation (U.S. patent publication US2006/0121605 by Parenteau, et al.) inadequate for cancer antigen identification for curative engineered T cell therapy, because it requires one to make assumptions regarding the applicability of the findings to a tumor's ability to grow and progress. Not all tumor types and stages within a particular cancer indication will have the same relationship between these compartments, yet this relationship will be important to tumor heterogeneity and progression.
The relationship between cell compartments within a tumor cell population cannot be appreciated by histopathology, genomics, proteomics or static marker analysis, if the analysis ultimately lacks the ability to link them to cell compartment interaction and behavior. Current methods have fallen short in their ability to distinguish between causative and contributing cell populations within a cancer type and stage. Evidence of this difficulty is the long running debate as to the clinical significance of cancer stem cells identified by these means.
Newer in vitro culture methods have been used to better understand the tumor cell population and its potential, but these methods are still inadequate to derive the key functional subpopulation responsible for tumor propagation and progression. For example, Gillet et al., PNAS, 108:18708 (2011)) determined that sixty established cancer cell lines (NCI60 panel) bear greater genomic resemblance to each other both in vivo and in vitro, regardless of tissue origin, than to primary tumors, indicating a need for a better way to maintain the fidelity of gene expression, epigenetic influences, cell relationships and cell response. This fidelity is particularly needed to discriminate robust yet safe peptide targets for CTL therapy. Using ovarian cancer as an example, in deriving of a tumor cell lines from ovarian cancers, Strauss et al. (Strauss et al. PLOSOne, 6:e16186 (2011)) found that ovarian cancer rapidly lost its epithelial component in culture, leaving a dominant yet seemingly hybrid epithelial/mesenchymal phenotype with an unclear relationship to the native tumor cells.
Although spheroid culture methods have been shown to maintain what are considered to be stem and progenitor cells (more likely primarily transit amplifying cells) within a heterogeneous population that can be expanded on a limited basis, the ability to dissect the cell populations from spheroids is difficult, requiring disaggregation of the spheres and clonal assays (Al-Hajj, et al. Proc. Nat. Acad. Sci. USA 100(7):3983-3988 (2003)), which then destroys the biological context. Also, evidence of selective regenerative stimulation of the progenitor pool is lacking in these methods when used to culture normal stem and progenitor cells (Parenteau, Cell Differentiation, In Vitro Mammalian. Encyclopedia of Industrial Biotechnology: Bioprocess, Bioseparation, and Cell Technology, John Wiley & Sons: 1-15 (2009)) and will not selectively stimulate the tumor's C-RC subpopulation, which is needed for functional identification. For example, derivation of a “cancerous” progenitor-like population in spheroid culture has required immortalizing normal ovarian epithelial cells with hTERT and causing partial transformation by overexpressing the oncogenes CMYC, KRAS or BRAF (Lawrenson, et al. Carcinogenesis 32(10):1540-1549 (2011)).
The gene signature from three-dimensional culture has been shown to provide prognostic value in breast cancers by comparing their genomic signatures (Martin, et al., PLoS ONE 3(8):e2994 (2008)). However like spheroid culture, it is incapable of functionally identifying and isolating the tumor C-RC subpopulation within the heterogeneous, differentiating tissue. Although 3-D culture serves as an valuable comparator, it favors differentiation and resulting heterogeneity, and does not selectively stimulate the regenerative tumor population. Therefore 3-D culture can be a valuable tool in assessing how a tumor develops and differentiates from the C-RC population but it does not isolate the C-RC population.
Animal models have served as an alternative way to analyze tumor biology in a more complex context. However animal implantation, while a functional test of the population, is often not a reliable estimate of tumor propagating cells (Maenhaut, et al. Carcinogenesis 31(2):149-58 (2010)). On one hand, the immune-compromised mouse can have some residual immunity, which can limit implantation and lower the tumor's ability to recruit cells of innate immunity that provide the stimulatory factors important for tumor development, in particular stromal and vascular recruitment (Shipitsin, et al., Lab. Invest. 88(5):459-463 (2008)). On the other hand, experience with tumorigenic cancer cell lines indicates that the development of tumors in mice from cultured cancer cells is a growth property that can be acquired over time, and need not relate to regeneration of the native tumor. Therefore tumor formation in animals in itself is insufficient to identify the implanted cells as the C-RC population of the native tumor i.e., the ability to form tumors in mice confirms a C-RC population of a tumor only if the cells used for implantation have been functionally identified as C-RC, prior to implantation. Using animal models as a tool to show recapitulation of the original tumor histology is important in demonstrating that the cells isolated are C-RC.
U.S. Pat. No. 8,309,354 discloses a method of deriving a population of cells containing cancer stem cells, by dissociating solid human tumors and culturing dissociated cells and small cell aggregates in serum-free conditions to obtain cancer stem cell lines that do not senesce upon serial passage, express certain stem cell markers that can vary between cancer types, and exhibit a high efficiency of tumor formation in mice. However the serum-free conditions used to generate these lines allow for the growth and persistence of a stromal cell component, which eventually gives way to a cancer stem cell-like epithelial cell population after an extended period (months) in culture. Similarly, Pan, et al. (Pan, et al. Methods 56:432 (2012)) describes deriving ovarian cancer stem cell-like lines where epithelial cells and stromal cells propagate in primary culture. A substantially pure population of epithelial cells can take 3-6 months, assuredly leading to substantial changes in the cancer cell population.
Recent advances in murine immune-compromised animal model development have given rise to several highly versatile strains expressing a variety of human immune markers in compartmentally valid patterns. These technologically advanced animals allow Human Leukocyte Antigen (HLA) matched and tissue specific expansion of selected tumor C-RC populations in vivo under conditions that enable their examination from both oncologic and immunologic perspectives (Covassin, Clin Exp Immunol 166(2):269-280 (2011)). In addition, this versatile mouse strain can be used as an in vivo incubator, as demonstrated by Bankert, et al. PLoS ONE 6(9):e24420 (2011) who showed that serous ovarian tumors could be successfully implanted and expanded in the NOD-scid IL2Rγnull (NSG) strain. However tumor heterogeneity is maintained and the tumors remain complex. Rather like 3-D organotypic models, use of these animals can serve as an important component in testing the robustness of expression and prioritization of C-RC-relevant tumor markers as well as potential C-RC behavior like epithelial-mesenchymal transition (EMT) and the associated potential for metastasis, once a C-RC population as well as ways to identify the population are in hand. In addition, these animals can be used for primary tumor expansion before in vitro derivation of the C-RC population with the added benefit for ACT in that TIL resident to the implanted tumor have been shown to be expanded as well. The ability to expand nascent TIL within these animal models provides the added opportunity to obtain valuable sources of tumor-specific T cells via standard antibody-based magnetic selection methods for TCR selection.
A method or technology for the derivation of a C-RC population of high fidelity and translational value for applications such as the targeting cancer cells for immune killing should be 1) inclusive, as C-RC-enabling mutation can arise from any dividing cell compartment within an epithelial lineage and identification based on human stem cell markers will be ambiguous and varied 2) response-based, for identification and selection, as only then can one be assured marker expression relates to the clinically relevant cell population and 3) high fidelity where technology does not artificially contribute to a shift in cell response or phenotype but permits a rapid, self-directed response through autocrine and paracrine signaling. Then, a method or technology useful for targeting a curative adoptive T cell therapy must be 1) capable of safely discriminating the C-RC population, i.e., those cells capable of continued tumor propagation and progression, and 2) able to link the expression of tumor markers to the C-RC population in vivo.
Despite many genomic, proteomic and biological attempts, ways to solve the conundrum of linking marker expression to the function of cells compartment within the tumor have not been obvious to researchers working to determine the biological significance of cancer markers. Despite genomics and proteomics yielding a large number of potential cancer markers (Polanski, et al. Biomarker insights 1:1-48 (2007)) and efforts to prioritize their therapeutic relevance (Cheever, et al. Clin Cancer Res. 15(17):5323-37 (2009)), researchers have had difficulty demonstrating their relevance as therapeutic targets. Researchers in the field recognize that this is due to the complexity of the tumor population and a lack of adequate systems to link the biology of a complex epithelial tumor to clinically applicable marker expression (Tysnes B B, Neoplasia 12(7):506-15 (2010)). Evidence of this deficiency is the continued debate as to the therapeutic relevance of cancer stem cells and in particular, the difficulty researchers have had in attempting to grapple with how to handle observed differences (Gupta, et al., Nat. Med. 15(9):1010-12 (2009); Maenhaut, et al. Carcinogenesis 31(2):149-58 (2010); Shipitsin, et al., Lab. Invest. 88(5):459-63 (2008); Pantie, J. Biosci. 36(5):957-61 (2011)) and ambiguous identifying markers (Jaggupilli and Elkord, Clinical and Developmental Immunology (2012), Article ID 708036).
It is clear that a functional way of deciphering cell capabilities within and between the cell compartments of a tumor sample is needed for practical and effective immunological targeting.
It is therefore an object of the present invention to provide a way of deciphering cell capabilities within and between the cell compartments of a tumor sample.
It is a further object of the present invention to focus a cancer's dynamic nature and variability by engendering a functional response to regenerative pressure in vitro so that key therapeutic targets can be identified.
It is a further object of the present invention to isolate the clinically relevant tumor cell population.
It is a further object of the present invention to use the clinically relevant population of a cancer type and stage as an antigen source to identify markers suitable for curative ACT using engineered T cells against that cancer type and stage.
It is a still further object of the present invention to use engineered T cells against markers that identify the clinically relevant tumor subpopulation to purposely design curative therapies.
It is another object of this invention to identify and validate the therapeutic significance of a putative cancer marker without the need for genomic screening, which can be prone to high levels of artifact and both positive and negative spurious findings.
It is a further object of the present invention to provide a means for deriving functional populations from a tumor and for using them to recapitulate tumor development in vitro and in vivo where bias and artifact can be identified, controlled and eliminated.
It is another object of the present invention to eliminate the need to presume the expression or clinical significance of a putative marker, such as a stem cell marker, in order to either identify or isolate the most clinically important tumor cell population.
It is another object of the present invention to provide methods and systems for programming T cells to selectively attack important tumor cells involved in proliferation, or invasion in an individual.
It is another object of the present invention to significantly reduce the risk and cost of development by functionally validating and prioritizing ACT targets prior to costly process development and clinical testing.